I find that a lot of people on LinkedIn take Gen AI too seriously, especially when they generate visuals. Not everything needs to be a boring corporate take (“OMG, not another quote on the wall generated with AI!”).
But prompting is inherently not fun. You need to be precise with words and be patient with iteration. In other words, you can’t “anyhowly” do it. I suspect this is why so much AI-generated content is so stuffy – the process overwhelms the intent.
So this is where you can learn from how creators blend creativity with skill – First, we swirl a fun idea in our heads. Then we apply our professional skills (eg. writing or drawing) which are honed through years of relentless practice. As we iterate, we focus on improving the idea, while reworking the output to be more precise and to fit the visual in our minds.
For Gen AI visuals, the skills you need are domain knowledge (vocabulary of visuals) and prompting knowledge (how to structure a prompt efficiently and precisely).
Here’s my final prompt for today’s satirical take on 2026 predictions. It took about 12 iterations to get the prompt right, and I used Nano Banana Pro in Google AI Studio to lock the aspect ratio. This method is pay-per-output and the total cost was about SGD $2.20.
Create a colorful and fun infographic in satirical colorful cartoon style like Mad Magazine (but without Mad Magazine characters). Asymmetrical layout. The numbers are big and bold, flushed next to the text. No other text or speech content. Fun font.
Headline: “5 Things AI Will Not Do In 2026”
“1. Eradicate pointless selfies”
“2. Write interesting posts”
“3. Get you to exercise more”
“4. Cure imposter syndrome”
“5. Give you hard skills”
Footnote: I could have just posted this image as a fun take on 2026, but then you don’t get any real value out of the $2.20 I spent!